Has anyone seen this issue before? I got it when using the AI build my pickaxe option and wanted to add some rules against prompt attacks.
You want to avoid adding security protocols using the AI builder. It’s best to use it to get a basic template that you can build off of.
After the AI helper is done laying the groundwork, add your security protocols at the end of your core prompt. That worked for me for every PA project.
I would just add that instruction manually to your bot’s prompt. The AI Builder thinks you are trying to do something suspicious and isn’t letting you do it.